Professional Documents
Culture Documents
Thermally Stable Iridium Contacts To P-Doped InGaAs
Thermally Stable Iridium Contacts To P-Doped InGaAs
Thermally Stable Iridium Contacts To P-Doped InGaAs
47 As for Indium
Phosphide Double Heterojunction Bipolar Transistors
Abstract
We report on surface pretreatment for ohmic contacts to p-doped In0.53 Ga0.47 As with improved thermal stability. It
is found that the cleaning of In0.53 Ga0.47 As surface by ammonium sulfide or sulfuric acid offers the optimum surface
treatment prior to metal deposition. Contacts using an iridium contact layer and palladium diffusion barrier were
fabricated and compared to a conventional platinum-based contact Pt/Ti/Pt/Au . Pt-based metal stack suffered from
void formation and high reactivity with the semiconductor when annealed at 240°C for a few hours, as examined by
transmission electron microscopy. As a result, the Pt-based stack exhibited strong deterioration of the resistivity. On the
other hand, the Ir contact maintained its integrity during thermal stress. The improved contact exhibited a void and
reaction-free microstructure and offered stable resistivity values with annealing.
2. Fabrication
6 0
30 nm thick. Even the removal of a thin layer of semi-
5 0 H 2 S O 4
conductor will increase the sheet resistance. Moreover, in
(N H )2S
4 0 H F
4
a graded doping base design, it is critical to preserve the
3 0 highly doped surface layer in order to minimize the tun-
2 0
neling barrier thickness.
In this experiment, medium voltage argon sputtering at
1 0
M e ta l s ta c k : P t/T i/P t/A u 1 0 /1 5 /1 5 /2 0 0 n m
90 V for 120 s was used with an Ar-gas load of 8.25 sccm
0 and an ion current of 15.6 A. The Ar+ bombardment was
Figure 3: Comparison of the measured as-deposited resistivity of the combined with a chemical pretreatment by H2 SO4 , con-
different surface treatment. Consistent testing was used for metal stack sidering that a thin layer of oxide readily forms while
and processing conditions. loading the sample into the vacuum chamber. Moreover,
reaching the required chamber pressure takes a couple of
In contrast, the samples treated with H2 SO4 and hours. This processing sequence has been established as
(NH4 )2 S emerged as a good compromise. Both solutions optimal on a similarly doped n-type In0.53 Ga0.47 As based
showed similar behavior, with relatively low resistivity, on previously carried out TLM measurements. On the p-
reasonably low sheet resistance and transfer length (Table In0.53 Ga0.47 As surface, however, this process introduced an
2). It is well documented in the literature [19, 20, 21] that extensive damage to the surface layer leading to higher
the sulfur passivation can effectively inhibit oxide growth. resistivity values. Therefore, it is desirable to have a de-
Sulfur atoms tend to form strong bonds with the InGaAs tailed understanding of the Ar+ bombardment etch rate
4
and the exact thickness of the p-In0.53 Ga0.47 As native oxide
before proceeding with this process in order to neutral-
ize its detrimental effects. In summary, the H2 SO4 and
(NH4 )2 S pretreatments offered an acceptable consump-
tion of In0.53 Ga0.47 As and an effective removal of native
oxides. The cleaning of the surface with Ar+ ions should
be used with caution, as it can introduce ion damage, re-
sulting in poor electrical performances.
4. Thermal Stability
Boosting the operation speed of DHBTs requires the
scaling of transistors for parasitic capacitance reduction
and increasing the current density for faster charging
times. Both approaches lead to significant heat genera-
tion inside the device and thus an instability at the metal-
semiconductor interface. In addition, the devices have to
withstand a considerable amount of heat exposure during
the BCB planarization processes and other thermal fabri-
cation steps, such as InP dry etching or plasma-enhanced
Figure 5: High-angle annular dark-field scanning transmission elec-
chemical vapor deposition of nitride layers. tron microscopy (HAADF STEM) micrograph of the Pt/Ti/Pt/Au
Thus, to evaluate the contact thermal stability, sam- 10/15/15/200 nm contact structure after annealing at 240°C for 7 hours.
ples with Pt/Ti/Pt/Au 10/15/15/200 nm contact to p- Voids appeared in the non reacted Pt layer, and strong intermixing is ob-
served between the In0.53 Ga0.47 As and Pt lower region.
In0.53 Ga0.47 As, cleaned with H2 SO4 were annealed for 7
hours at 240°C under constant flow of N2 . The first layer
of platinum is the contact layer, the thin Ti layer acts as an
adhesion film and the second Pt is the diffusion barrier. semiconductor interface will be pushed to a lower-doped
Following the annealing process, a microstructural in- region. As the contact is exposed to more heat, the plat-
vestigation using TEM revealed two major phenomena: inum continues to sink into the InGaAs, consuming more
extensive void formation in the nearly unreacted region semiconductor material and leading to severe degrada-
of the platinum layer as shown in Figure 5, and deep pen- tion of the tunneling proprieties due to the wider space-
etration of the platinum into the semiconductor. The voids charge extension. This adds further metallurgical insta-
are thought to be caused by a Pt flux toward the lower re- bility and compromises yield and reliability of the device.
action region that is not being balanced by a correspond- Figure 6 illustrates the EDX depth profile of a Pt/Ti/Pt/Au
ing flux of group III and group V elements in the opposite metal stack after annealing for 7 hours. The platinum
direction [11]. No oxide layer was detected, confirming intensity does not drop sharply at the interface with the
previous studies that Pt disperses the native oxide in III-V In0.53 Ga0.47 As and the intermixing between the layers ex-
semiconductors during annealing [11, 10, 26, 27]. The tends over several nanometers.
decomposition of the native oxide could be a contribu- The process produces a Pt-In-Ga-As sublayer and si-
tor to the formation of voids, although this theory would multaneously depletes the Pt film as a result of the out-
have to be examined more thoroughly. Throughout the diffusion, resulting in void formation and the intermixing.
experiment, the semiconductor sheet resistance between The second reaction region was not conclusively identi-
the structures remained fairly constant suggesting that the fied, although other studies speculate it may consist of the
heat treatment has minimal impact on the carrier density following intermetallics : Ga3 Pt5 , GaPt, In2 Pt, InAs, InTi3 ,
at the exposed semiconductor surface. This result also in- Ti2 Ga3 , and GaPt3 [28, 29]. Further exposure to the ele-
dicates that the metal diffusion in the semiconductor was vated temperature results in increased vertical spreading
mainly vertical. The SiNx passivation effectively protected of the sub-layers, leading ultimately to the failure of the Pt
the gap region from charge related degradation, but did diffusion barrier and the penetration of Au into the semi-
not thermally isolate the contact. The voids in the metal conductor.
stack, however, represent a serious mechanical degrada- Therefore, shallow contacts with stable morphology are
tion, an issue to be addressed in order to ensure long-term required. Refractory metals such as W, Mo, and Ir, are
device reliability. good candidates for a new generation of contacts. These
The second mechanism responsible for the deterioration metals have the advantage of lower diffusivity through
of the contact is the diffusion of the platinum and its re- the semiconductor at high temperature and offer a higher
action with In0.53 Ga0.47 As. During annealing, the platinum resistance to chemical reaction with the semiconductor
moves deeply into the reaction region. Because of the ex- when exposed to thermal processing. In addition, they
istence of a doping gradient in the base layer, the metal- are extremely resistant to corrosion and heat due to their
5
analysis using TEM was performed on the Ir sample.
As observed in figure 7, the contact shows a chemically
abrupt interface. Electron diffraction experiment did not
reveal any presence of intermixing compounds showing
only textured Ir layers (Figure 8). The out-diffusion ap-
pears to be suppressed effectively. No voids in the upper
layers were observed, even when annealed for 7 hours,
demonstrating excellent microstructual stability.
Figure 6: EDXS line profile showing elemental distribution across the Pt-
based metal stack and metal-semiconductor interface after aging at 240
°C for 7 hours.
2 0 0
1 5 0
1 0 0
5 0
0 1 3 7
A n n e a lin g tim e / h
Figure 7: Cross-sectional HAADF STEM micrographs showing Ir-based Figure 9: Comparison between the measured contact resistivity of the
metal stack on InGaAs after annealing at 240°C for 7 hours, b) a higher Pt-based and Ir-based metalization scheme as a function of annealing
magnification image of Ir/InGaAs interface. Metal diffusion into the time. results are averaged over many samples processed using the same
semiconductor material was prevented by the Ir contact layer. conditions.
After annealing for 7 hours, a metallurgical stability As expected with the conventional metal stack, strong
6
degradation occurs due to the reason mentioned above, Conclusion
the Ir stack in contrast exhibited small deviations from the
as-deposited values, thus supporting the expectation that In this report, the influence of surface treatment and
field emission is the prevailing current transport mecha- stability of multilayer ohmic contacts to p-type InGaAs
nism due to the limited reaction between the semiconduc- were examined. The H2 SO4 and (NH4 )2 S pretreatment
tor and the refractory metal. offered an acceptable consumption of InGaAs and an ef-
fective removal of native oxides. It was also found that Pt-
Time As dep. 1h 3h 7h based metal exhibit low long-term reliability under ther-
mal stress, as they suffer from high reactivity along with
Pt Ir Pt Ir Pt Ir Pt Ir
a deep penetration into the semiconductor when exposed
ρC [Ω.µm ] 2
40.1 34.4 108 41.9 210 44.5 249 48 to N2 atmosphere at 240°C over several hours. The an-
RSH [Ω/] 710 769 650 788 564 787 536 785 nealing process results in the formation of Pt-In-Ga-As in-
RC [Ω.mm] 0.18 0.16 0.26 0.18 0.34 0.18 0.36 0.19 terlayers and simultaneously depletes the Pt film leading
Lt [µm] 0.24 0.21 0.4 0.23 0.6 0.23 0.68 0.24 to an observed void formation. This results in a resistiv-
ity increase from 40 Ω.µm2 to 240 Ω.µm2 , as the current
transport mechanism is severely deteriorated. A novel Ir-
Table 3: Summarized measurement results of the contact resistivity with based metal stack was successfully demonstrated. TEM
annealing temperature for the Pt and Ir contacts. All samples were
treated with the H2 SO4 and processed under the same conditions.
analysis showed that iridium can effectively inhibit inter-
mixing due to its strong resistance to reactions with met-
On the other hand, The resistivity of the Iridium-based als or semiconductors. The new contact maintained its
contact remained practically unchanged throughout the integrity during thermal stress, yielding a stable metallur-
annealing process. With variation less then 20% for ρC gical interface and low resistivity values.
after 7h of annealing at 240°C, the Ir clearly presents su-
perior thermal stability compared to the conventional Pt
Acknowledgment
contact. The results are summarized in table 3. The ab-
sence of reactions at the Ir/InGaAs interface is further con-
The authors would like to thank the FBH process tech-
firmed by the stable electrical measurement of the sheet
nology department, especially A. Dounia and J. Cwynar
resistance.
for chip measurement, N. Thiele and A. Runge for the
S tre s s c o n d itio n s : 9 h , N 2 4 0 ° C
3 0 metal deposition, B. Janke, M. Matalla for photomask cre-
2
1 0 -2
P t-b a s e d c o n ta c t
ation and U. Zeimer for the constructive discussions. This
Ir-b a s e d c o n ta c t 2 5
work has been partially supported by the Horizon 2020
research and innovation program under grant agreement
-3 IC 2 0
1 0
no. 762119 (ULTRAWAVE) and the Leibniz Gemeinschaft
C u r r e n t g a in
1 5
-4
1 0
β 1 0
IB References
-5 5
1 0
[1] M. Urteaga et al., InP HBT Technologies for THz Integrated Circuits,
Proceedings of the IEEE Vol. 105, No. 6, June 2017
0 [2] J. C. Rode et al, Indium Phosphide Heterobipolar Transistor Technol-
0 .4 0 .5 0 .6 0 .7 0 .8 0 .9
ogy Beyond 1-THz Bandwidth, IEEE Transactions On Electron De-
V B E / V vices, Vol. 62, No. 9, September 2015
[3] E. F. Chor et al., Metallurgical Stability of Ohmic Contacts in Base
Figure 10: Comparison of the Gummel plots of two large area devices InP/InGaAs/InP HBT’s, IEEE Electron Device Letters, Vol 17, No. 2
(LAD) transistors having a different base metalization after aging. The Ir February 1996.
contact exhibited a higher current gain and lower base current. [4] J. Rode et al., An InGaAs/InP DHBT With Simultaneous fτ /fmax
404/901 GHz and 4.3 V Breakdown Voltage, Journal of Eletron De-
vice Society., 2014
The observed deterioration in the transport mechanism [5] N.G. Weimann et al., SciFab – a wafer-level heterointegrated InP
can be directly reflected in the DC characteristics of the DHBT/SiGe BiCMOS foundry process for mm-wave applications,
large area transistors, as shown in figure 10. The parasitic Phys. Status Solidi A, 1–8 (2016)
base current is significantly higher with a Pt contact than [6] H. Masuda et al., Device Technology of InP/InGaAs HBTs for 40-Gb/s
Optical Transmission Application, GaAs IC Symposium. IEEE Gal-
for the Ir-based device. The results indicate degradation lium Arsenide Integrated Circuit Symposium. 19th Annual Techni-
in the crystal quality beneath the Pt contact, generation of cal Digest 1997.
defects at the interface is responsible for increased recom- [7] E. F. Chor et al. Electrical characterization, metallurgical investiga-
tion, and thermal stability studies of Pd, Ti, Au-based ohmic contacts,
bination and trap states [30, 31]. As a result, a higher base Journal of Applied physics 2000, Volume 87, Number 5.
current and base ideality factor are observed for the con- [8] J. C. Lin, Characterization of low-resistance ohmic contacts to n- and
ventional contact, along with a drop in the current gain. p-type InGaAs, Journal of Applied Physics 114, 044504 (2013)
7
[9] J.S Yu et al., Pt/Ti/Pt/Au and Pd/Ti/Pt/Au ohmic contacts to p-
InGaAs, Compound Semiconductors 1997, Proceedings of the IEEE
Twenty-Fourth International Symposium on Compound Semicon-
ductors.
[10] J. A.Robinson et al., Solid-State Phase Formation between Pd Thin
Films and GaSb, Journal of Electronic Materials , Vol. 35, No. 1,
2006
[11] R. Dormaier Thermal stability of Pd/Pt/Au Ohmic contacts to In-
AlSb/InAs heterostructures for HEMT, Journal of Applied Physics
105, 044505 (2009);
[12] Baraskar, Development of Ultra-Low Resistance Ohmic Contacts for
InGaAs/InP HBTs, UCSB 2011 Doctoral Dissertation.
[13] E. Lobisser et al., Ex-situ Tungsten Refractory Ohmic Contacts to p-
InGaAs, 38th International Symposium on Compound Semiconduc-
tors, Berlin 2011
[14] H.H. Berger, Model for contacts to planar devices,Solid States Elec-
tronics, Vol. 15 pp. 145-158, 1972
[15] F.L. Lie et al., In0.53 Ga0.47 As (100) native oxide removal by Liquid
and Gas phase HF/H2O chemistry, Microelectronic Engineering 87
: 1656-1660 November 2010
[16] J. Bardeen, Surface states and rectification at a metal-semiconductor
contact, Physical review Volume 71 Number 10, 1947
[17] G. Strateev et al., A controllable mechanism of forming extremely
low-resistance non-alloyed ohmic contacts to III-V semiconductors, J.
App. Phys. 74 (12) 1993
[18] V. L. Rideout, A review of the theory and technology for ohmic con-
tacts to group III-V compound semiconductor, Solid state electronic
Vol 18 pp 541-550. 1975
[19] T.P. Chen et al, Surface treatment effect on temperature dependent
properties of InGaP/GaAs heterobipolar transistors, Journal of Ap-
plied Physics 101, 034501 (2007).
[20] R. Driad et al, Passivation of InGaAs surfaces and InGaAs/InP het-
erojunction bipolar transistors by sulfur treatment, Appl. Phys. Lett.
73, 665 (1998)
[21] H. Wang et al, Design Of Shallow And Thermally Stable Contacts On
Antimonide-Based Semiconductors, Doctoral dissertation, The Penn-
sylvania State University, 2004
[22] X.A. Cao et al., Effects of interfacial oxides on Schottky barrier con-
tacts to n- and p-type GaN, Applied Physics Letter, Volume 75 Num-
ber 26 (1999)
[23] M. Diale et al., Effects ofchemical treatment on barrier height and
ideality factors of Au/GaN Schottky diodes, Physica B: Condensed
Matter Volume 404, Issue 22, 1 December 2009, Pages 4415-4418
[24] G. Stareev et al., A Reliable Fabrication Technique for Very Low Resis-
tance Ohmic Contacts to p-InGaAs Using Low Energy Ar+ Ion Beam
Sputtering, Proceedings 1991 of Third International Conference In-
dium Phosphide and Related Materials.
[25] M. Alexandrova et al., Development and Optimization of High-Speed
InP/GaAsSb Double Heterojunction Bipolar Transistors, Dissertation.
ETH Zürich No. 23197, 2015
[26] T. Sands et al., Ni, Pd, and Pt on GaAs: A comparative study of
interfacial structures, compositions, and reacted film morphologies,
Journal of Material Research, Volume 2, Issue 2 April 1987 , pp.
262-275
[27] J.A. Robinson et al., Pd/Pt/Au ohmic contact for AlSb/InAs0.7Sb0.3
heterostructures, Solid-State Electronics Volume 50, Issue 3, March
2006, Pages 429-432
[28] D. G. Ivey et al, Microstructural analysis of Au/Pt/Ti contacts top-
type InGaAs, Journal Of Materials Science: Materials In Electronics
6 (1995) 219-227
[29] A. Katz et al, Pt/Ti ohmic contacts to ultrahigh carbon-doped p-
GaAs formed by rapid thermal processing, Appl. Phys. Lett. 56, 1028
(1998)
[30] Yoshino K. Fukai and Kenji Kurishima, Reliability Study of InP-
Based HBTs Operating at High Current Density, Material and Reli-
ability Handbook for Semiconductor Optical and Electron Devices,
Springer Science and Business Media New York 2013
[31] K.T. Feng et al, Investigation of reliability for C-doped
InP/InGaAs/InP HBTs under high current density operation,
Proceedings of GaAs Reliability Workshop, 2003, pp. 117–120.